Notices
Results 1 to 5 of 5

Thread: Evidence Based Medicine

  1. #1 Evidence Based Medicine 
    New Member
    Join Date
    May 2013
    Posts
    3
    I am interested in the broad subject of "evidence based medicine" where the word "evidence" seems misused. Much of evidence based medicine stands on null hypotheses, test statistics, arbitrary alpha levels, P-values, and often arbitrary rulings concerning "statistical significance" (or not). These are old concepts and poor at best in the 21st century. Medical research is very well funded; why are these old methods still in use (they have been called "paleostatistics").

    My work over the past 20+ years has been on a different paradigm -- called information-theoretic. This approach is very different from the historical methods: there is no test, no alpha level, no P-value, no ruling of "significance." These methods avoid the "priors" required in the Bayesian approach. The new approaches allow information loss to be quantified. This leads to inferential quantities such as the likelihood of model i given the data and the probability of model i given the data. An evidence ratio is merely the ratio of these quantities. More importantly it is easy to base formal inference on all the models in a set -- called "multimodel inference." The 21st century will be all about multimodel inference.

    P-values are not evidential! This is well known among statisticians. In fact, statisticians have been critical of P-values since the last 1930s. I think very few statisticians belive in or use P-values in serious research. In fact, very few people using P-values can offer a correct definition!

    I would like to help people change from using the old methods to modern approaches. Medical reports often conclude "no evidence, P = 0.07, when alpha is 0.05. This is wrong/stupid; of course there is evidence in this example (the Supreme Court has just ruled on this matter).

    Are people interested in discussion these issues?

    David Anderson


    Reply With Quote  
     

  2.  
     

  3. #2  
    Forum Masters Degree LuciDreaming's Avatar
    Join Date
    May 2013
    Location
    Cambridgeshire
    Posts
    656
    Is this Claude Shannons Information theory you are talking about?


    "And we should consider every day lost on which we have not danced at least once. And we should call every truth false which was not accompanied by at least one laugh" Nietzsche.
    Reply With Quote  
     

  4. #3  
    New Member
    Join Date
    May 2013
    Posts
    3
    No, it is Kullback-Leibler information. This is an extension of Shannon information. It is sometimes called cross entropy.

    30 years ago a Japanese statistician found a like between K-L information and Fisher's loglikelihood function and this opened the door to many useful inferential methods.

    David Anderson
    Reply With Quote  
     

  5. #4  
    Moderator Moderator
    Join Date
    Nov 2011
    Location
    city of wine and roses
    Posts
    6,222
    For the time being, it doesn't matter if there's a more elegant technical solution to these problems. Whatever system is proposed to replace the current one, someone has to persuade all the universities and medical schools, all the funding bodies, all the governments, all the pharmaceutical companies and all the international medical professional associations that the proposed way is better, easier to understand and cheaper than the current system and all proposed variants, improvements or replacements.

    Good luck with that.
    "Courage is what it takes to stand up and speak; courage is also what it takes to sit down and listen." Winston Churchill
    "nature is like a game of Jenga; you never know which brick you pull out will cause the whole stack to collapse" Lucy Cooke
    Reply With Quote  
     

  6. #5  
    New Member
    Join Date
    May 2013
    Posts
    3
    Yes, I hear you. Sometimes progress rides in a hearse. However, inferences are routinely being made that are simply incorrect; that should worry professional people.
    Thanks

    David
    Reply With Quote  
     

Similar Threads

  1. Evidence and no evidence of time travelers
    By kelleskurter in forum Science-Fiction and Non-Fiction
    Replies: 20
    Last Post: November 23rd, 2013, 05:17 AM
  2. Can science use evidence of evidence?
    By grandi in forum General Discussion
    Replies: 6
    Last Post: September 20th, 2012, 06:03 PM
  3. Earth-Based Versus Space-Based Telescopes
    By Jagella in forum Astronomy & Cosmology
    Replies: 10
    Last Post: October 17th, 2011, 01:03 AM
  4. A Knowledge-Based Bailout Plan Based On Solution Of Value
    By jackson33 in forum Business & Economics
    Replies: 0
    Last Post: October 8th, 2008, 11:53 AM
  5. Objective evidence vs Anecdotal evidence
    By arkofnoah in forum Scientific Study of Religion
    Replies: 3
    Last Post: September 21st, 2008, 10:39 AM
Bookmarks
Bookmarks
Posting Permissions
  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •